Digital Diogenes: Search and Browse


February 2003

Digital Diogenes: Search and Browse

CIDMIconNewsletter David Walske, David Walske, Inc.

The Analog Closet

Several years ago while attending a popular online Help conference, I sat next to a fellow presenter whom I knew. As I opened my ever-present, paper-based agenda and began to take notes, he chided jestingly, “Looking pretty analog there, buddy.” His comment had more of an effect on me than he might have expected. At my very next opportunity, I bought a PDA and began migrating my personal and professional life into its digital format. Although the transition was far from painless, I persevered and prevailed. Day after day, I synchronized contact, task, and calendar information between handheld and PC-based applications.

One year later, I gave it up and revived my old favorite leather-bound, paper-based agenda, migrating my life back to analog. And you know what? I felt like I was home again. I do still manage certain regular repeating events, as well as my contact list, with PC and handheld. But for the rest, I’m happy to have the paper back in my hands.

Chaos controlled or controlled chaos?
In the grand scheme of things, how I choose to manage my appointments is a small matter. But few would contest air traffic control as the archetypal mission-critical application. Through the years, many of our favorite movies have dramatized this pressure-cooker process; the dark, smoky rooms perched high above airport runways, crowded with air traffic controllers hunched over radar consoles.

Looks like I picked the wrong week to stop smoking.

-Lloyd Bridges, 1913-1998, in the film Airplane

Some airlines even allow curious passengers to listen in on the air traffic com-chatter-check out audio channel nine on your next flight. The allure is undeniable. Very hi-tech stuff, this air traffic control business. Techno-drama no less, as air traffic controllers each track up to several dozen planes as if they were so many flying donuts, each traveling at various speeds, different altitudes, and toward many destinations.

You might be surprised to learn that, according to Malcolm Gladwell in “The Social Life of Paper” in The New Yorker Magazine, March 25, 2002, “…as a controller juggles all those planes overhead, he scribbles notes on little pieces of paper, moving them around on his desk as he does.” Detractors of our less than perfect air traffic control system are quick to point out the folly in this paper chase. But the truth is that at the human level of interaction, this origami approach to content management appears to work quite well. Air traffic controllers successfully blend digital information with the tactile control added by the scraps of paper.

Look around your office. Do you have more or less paper surrounding you than you did ten years ago? If you’re like most of us, the answer is more. What is the meaning of all this paper in the so-called paperless office? What is the point of all our hard drives and servers storing mega-, giga-, and terabytes of data? Take another look around your office. See any filing cabinets? You might see a few. Ten years ago, filing cabinets dominated the business environment. Today, they’re almost a decoration, an afterthought (see Figure 1).

Hard disk drives condense a room full of filing cabinets into a few square inches. As advancing technology gives us the ability to jam more and more information into a smaller and smaller physical space, we begin to create the virtual equivalent of a black hole. Throw a chunk of matter into a black hole, and it is gone forever. Toss a chunk of content into a database, and you risk losing contact with it.


Figure 1. The Author, Circa 1980, Working Amid the Filing Cabinets


Paper is clearly still with us, in greater quantity than ever. No longer the prima facie choice for long-term storage, paper is a preferred output medium. When a Web page really grabs your attention what is your first impulse? Click Print. Even the mighty e-book cannot claim victory over paper. Nothing beats the feel of a favorite book in your hands. Indeed, sometimes paper not only covers rock but silicon as well.

It is clear that the choice of object that is one of the elements in the harmony of form must be decided only by a corresponding vibration in the human soul.

-Vassily Kandinsky, 1866-1944

Vassily Kandinsky is widely considered to be the inventor of 20th century abstract painting.

We have an affinity for paper. Paper is tactile. Jef Raskin, in his book The Humane Interface (Addison-Wesley 2000), describes a project planning room-a physical room in an office building not a “chat” room-in which “…the walls become covered with annotations, tacked-up sheets of paper, sticky notes, photos, or whatever helps us to remember or explain our ideas.” (page 153) The papier mâché wallpaper not only helps us to remember our ideas, it also helps us more effectively organize the individual chunks of content represented by the sheets and scraps of paper into an information architecture that serves a larger content deliverable.

Content management is about the effective storage, retrieval, assembly, and reuse of very large collections of information. Searching and browsing is about finding content: retrieval. Finding the right chunks of information requires an affinity for the content, such as that provided by the paper-covered walls of Raskin’s project planning room. Converting terabytes of information onto sheets of paper might rival “Christo’s Umbrellas” but would not be an effective component of content management (see the sidebar).

A humane interface
Where is the bridge upon which we may safely cross the digital divide? Raskin and his colleagues have proposed a zooming interface paradigm (ZIP) for user interface (UI) designs. In this paradigm, the user zooms effortlessly between the 10,000-foot overview and the microscopic view. Entering a physical space, say a public library, we survey the room before zooming in on a subject area, then on a specific book and a particular passage. When we need to access other books or other subject areas, we zoom out. The ZIP interface mimics natural navigation patterns.

Most search and browse UIs suffer from two fatal flaws. First, most rely upon a one-dimensional index of metadata. The Dewey Decimal Classification (DDC) system is an indexing system based on one view of human knowledge. Dewey is regarded as the father of modern library science, but his view of information may not be yours.

…it’s pretty hard to be excited about a whole curriculum of facts

[about which] everyone is worried that everyone needs to know. The reality is, there is no one set of things that everyone needs to know… but there are a set of things we know about the mind, an open, expanding learning… environment that is happily going off trying to find new things.

-Roger Schank, Cognitive psychologist

The second flaw of most search and browse UIs is that they do a pretty good job of getting in between you and the content. To find a desired chunk of content, you spend time wading through layers of UI and very little time with the content. Navigating an interface is the opposite of the tactile contact we enjoy with paper.

Glass Engine

In early 2000, composer Philip Glass was pondering the task of putting his catalog of music on the Web. The Glass catalog is a collection of musical compositions for film, theatre, and dance, as well as a large body of concert works. Certainly a database with content of such richness and depth would require a superior navigational UI. Glass turned to Mark Podlaseck of IBM’s Watson Research Center.

Spectral navigation
Mark began by examining the navigational tools of existing Web-based musical catalogs. What he found was not very promising. Web-based tools for searching music libraries largely forced the user to conform to a predetermined information architecture.

In this approach to navigation, the user is separated from the content by multiple UI layers. To move through the layers, the user must make a series of decisions. Each decision defines an increasingly restricted view of the database. A wrong decision anywhere in the hierarchy causes the process to fail.

Often, such content navigation is handicapped by search and browse algorithms that treat semantic metadata as true or false. For example, in Figure 2 the metadata for Classical is either on or off. Instead, Podlaseck envisioned a navigation that treated the semantic metadata as an analog spectrum. A piece of music could easily overlap Classical and Popular. Such a content chunk might be 60 percent classical and 40 percent pop.


Figure 2. A Typical Web Search and Browse Hierarchy

Concurrent spectra
Mark established a set of objectives for the Glass Engine project:

  • Provide the ability to view the entire Philip Glass catalog from many perspectives at once.
  • Express affinities without commitment or penalty.
  • Maximize exposure to music with rapid preview.
  • Iteratively refine or personalize the view of the catalog.

These goals mirror our natural processes. As we delve into a database, we do so with a unique perspective and set of expectations. The Glass Engine project objectives require a highly flexible taxonomy that adapts to the user’s expectations. A single spectrum of navigation, while an improvement, is still not enough. Glass Engine would require several linked, concurrent spectra (see Figure 3).


Figure 3. Glass Engine

Track selections are made by using any of several navigation bars to select a particular piece of music. Expressing affinities for a musical attribute does not restrict the catalog view. All the tracks are always available. Once an affinity is expressed by moving a navigation bar, the selected track plays immediately. There are no blind alleys and no layers of hierarchy to wade through. The catalog selections are immediately accessible. Very quickly, the UI becomes transparent.

Glass Engine uses an XML Schema to map data attributes to its UI. Advanced filtering, bookmarking, and complex query are included for users that require them. But the outstanding feature of Glass Engine is just how quickly new users feel at home and are able to navigate through the catalog.

Keys to Successful Content Management

In this article, we’ve explored some examples of good interface design and intelligent use of metadata for content retrieval. Imagine such intelligence in a content-management system (CMS) used for information development in a collaborative single-sourcing environment. Remember, your information developers can’t reuse content if they can’t find it in your database. There are many considerations in planning and implementing a content-management system. Providing easy-to-use, effective, information-retrieval tools is one of the key tasks of successful content management. CIDMIconNewsletter

About the Author

February BP23